A multiblocked mutant strain (ΔAHBA and Δasm12, asm21) of Actinosynnema pretiosum, the producer of the highly toxic maytansinoid ansamitocin, has been used for the mutasynthetic production of new proansamitocin derivatives. The use of mutant strains that are blocked in the biosynthesis of an early building block as well as in the expression of two tailoring enzymes broadens the scope of chemo-biosynthetic access to new maytansinoids. Remarkably, a ring-expanded macrolactone derived from ansamitocin was created for the first time.
A multiblocked mutant strain (ΔAHBA and Δasm12, asm21) of Actinosynnema pretiosum, the producer of the highly toxic maytansinoid ansamitocin, has been used for the mutasynthetic production of new proansamitocin derivatives. The use of mutant strains that are blocked in the biosynthesis of an early building block as well as in the expression of two tailoring enzymes broadens the scope of chemo-biosynthetic access to new maytansinoids. Remarkably, a ring-expanded macrolactone derived from ansamitocin was created for the first time.
A promising strategy for creating natural product libraries is to combine biotechnologically oriented synthetic biology1 with synthetic chemistry.2 The former allows complex molecular architectures to be created through multistep biosynthesis while the latter provides large synthetic flexibility for introducing unnatural but pharmacologically important functions and functional groups.Sensibly, chemo‐biosynthetic strategies can be categorized by dissecting a synthesis into chemo‐ and biosynthetic elements (Figure 1). Semisynthesis is an example for a BIO‐CHEM strategy (case A1) in which first a natural product is formed naturally followed by chemical derivatization.2b If the biosynthesis is blocked at a late stage the organism can provide advanced biosynthetic intermediates derived from important secondary metabolites (Figure 1, case A2).2c
Figure 1
Selected concepts of chemo‐biosynthetic strategies (blue: chemical synthesis, orange: biosynthesis, dark blue: final product, green: position of block; semisynthesis=BIO‐CHEM and three concepts of mutasynthesis= CHEM‐BIO; α, β, γ=enzymes; a, b=chemical agents/catalysts).
Selected concepts of chemo‐biosynthetic strategies (blue: chemical synthesis, orange: biosynthesis, dark blue: final product, green: position of block; semisynthesis=BIO‐CHEM and three concepts of mutasynthesis= CHEM‐BIO; α, β, γ=enzymes; a, b=chemical agents/catalysts).An example for the reversed CHEM‐BIO scenario (case B) is mutational biosynthesis (short “mutasynthesis“) in which for example a mutant strain of a producer organism is blocked in the formation of an early biosynthetic pathway intermediate.3 Administration of so called mutasynthons that are commonly prepared by chemical synthesis – these are structural analogs of biosynthetic intermediates – to the mutant strain then results in new metabolites sometimes referred to as mutaproducts (Figure 1, case B1). So far, no reports have appeared that utilize mutant producer strains that are blocked both at an early as well as at a late stage of the biosynthesis (Figure 1, case B2).Following the concept B2, we disclose the creation and mutasynthetic exploitation of a mutant strain of Actinosynnema pretiosum, the producing strain of the highly cytotoxic polyketide ansamitocins P‐3 (3), a maytansinoid.4 Only in 2013 it was approved by the FDA as an antibody drug conjugate against HER2‐positive breast cancer.5 The strain is genetically blocked in the formation of 3‐amino‐5‐hydroxybenzoic acid (AHBA, 1) as well as in two tailoring steps, namely the chlorination (Asm12) and the amidination (Asm21) leading to almost complete loss of post‐PKS enzyme activity (Scheme 1).6
Scheme 1
Overview of ansamitocin 3 biosynthesis via proansamitocin (2), genes of tailoring enzymes and genes blocked in new strain AH1 of Actinosynnema pretiosum; (green: position of blocks in this work).6, 10
Overview of ansamitocin 3 biosynthesis via proansamitocin (2), genes of tailoring enzymes and genes blocked in new strain AH1 of Actinosynnema pretiosum; (green: position of blocks in this work).6, 10As a consequence, new proansamitocin derivatives can be obtained that can principally be further modified semisynthetically thereby further expanding the opportunities to generate new nature‐derived macrolactams.Over the past years7 we mostly employed ΔAHBA mutant strains (resembling case B1). AHBA 1
8 is formed from UDP‐glucose through a unique variation of the shikimate acid pathway (Scheme 1).9 Processing through seven PKS modules furnishes seco‐proansamitocin that undergoes macrolactamization by an amide synthase that is not part of the PKS to yield proansamitocin (2).10, 11 Finally, six tailoring steps provide ansamitocin P‐3 (3).10 Commonly, the protocols can be conducted with unmodified benzoic acids, because activation to the corresponding SCoA esters occurs before the in vivo uploading onto the PKS.12The new mutant strain (AH1) was constructed according to the procedure of Yu et al.10 The plasmid pHGF9029 was delivered into A. pretiosum Δasm12/21
6 by conjugation with Escherichia coli. Positive clones where detected by feeding experiments with and without AHBA 1. Fermentation conditions were adapted to those previously described for the ΔAHBA mutant strain (HGF073) of A. pretiosum.7a The monitoring of proansamitocin titers during fermentation revealed that highest concentrations are observed on day ten post inoculation (p.i.). For ansamitocin production13 the addition of the amino acid L‐valine was beneficial. Unexpectedly, also in the present case increased proansamitocin production was observed upon supplementation of L‐valine. In fact, L‐valine is a biosynthetic precursor for the ester side chain at C3 and this transfer does not occur in strain AH1. We also found that pulse feeding of 2.5 g/L fructose on days 3, 4 and 5 p.i. led to significantly increased proansamitocin (2) titers (Figure S2 in the Supporting Information) which was also reported for ansamitocin P‐3 (3) titers produced by the wild‐type strain.14Upon feeding of AHBA 1 to strain AH1 proansamitocin (2) and several derivatives 2, 4–11 were isolated after extensive HPLC. As expected, proansamitocin (2) was the main product formed (Figure 2). Macrolactams 4–6 were disclosed before when using a Δasm12/21 mutant strain of A. pretiosum.15 19‐Hydroxy proansamitocin (7), 10‐desmethoxy proansamitocin (8) and macrolactams 9–11 that mainly differ in the stereochemistry and location around the diene moiety (C10–C14) are new members of the proansamitocin family. The structure elucidation relied on NMR spectroscopic analyses that included the determination of nuclear Overhauser effects (nOe; Figure 3).
Figure 2
Proansamitocin (2) and (oxygenated) derivatives 4–11 produced by A. pretiosum (strain AH1) after supplementation of AHBA 1. [a] Stereochemistry at C10 not assigned.
Figure 3
Coupling constants of 10‐H in (10R)‐2 and (10S)‐proansamitocin 4, 2D NOESY correlations in 1,3‐diene 8 and selected 1D nOe correlations in 1,4‐diene 11.
Proansamitocin (2) and (oxygenated) derivatives 4–11 produced by A. pretiosum (strain AH1) after supplementation of AHBA 1. [a] Stereochemistry at C10 not assigned.Coupling constants of 10‐H in (10R)‐2 and (10S)‐proansamitocin 4, 2D NOESY correlations in 1,3‐diene 8 and selected 1D nOe correlations in 1,4‐diene 11.Having established the ability of the new multiblocked mutant strain AH1 to accept and process AHBA 1, we next administered different aromatic acids 12–31 to a growing culture of strain AH1 (scale 3 x 50 mL). The formation and presence of new proansamitocin derivatives was determined by UPLC‐HRMS/MS analyses after extraction with ethyl acetate.Halogenated 3‐aminobenzoic acids 12–16 were processed by strain AH1 to the corresponding macrolactams (Figure 4). Mass spectrometric analyses of the crude extracts obtained after feeding of 3‐amino‐(het)arenecarboxylic acids 17–22 revealed no acceptance and processing to the corresponding proansamtocin derivatives.
Figure 4
Aromatic acids 12–31 tested as mutasynthons for strain AH1. First MS data (m+Na+) of expected proansamtocin derivatives collected from crude products are given. [a] No PKS‐processed products detected; [b] detoxification products (benzamides) or starting benzoic acids detected; [c] processing occurred but only seco‐acids detected. Test fermentations were carried out in three parallel experiments accompanied by two positive controls (with AHBA addition) and one negative control (without addition of mutasynthon).
Aromatic acids 12–31 tested as mutasynthons for strain AH1. First MS data (m+Na+) of expected proansamtocin derivatives collected from crude products are given. [a] No PKS‐processed products detected; [b] detoxification products (benzamides) or starting benzoic acids detected; [c] processing occurred but only seco‐acids detected. Test fermentations were carried out in three parallel experiments accompanied by two positive controls (with AHBA addition) and one negative control (without addition of mutasynthon).A second series of benzoic acids that are not anilines but rather 3‐aminomethyl‐ (23) and 3‐hydroxymethylbenzoic acids (24, 25 and 27–31) as well as phenylacetic acid derivative 26. HRMS analysis of crude extracts obtained from feeding experiments conducted with 24 and 25 revealed the presence of macrocyclic products while the corresponding seco‐acid was the result from mutasynthetic processing of 23.Surprisingly, feeding experiments with phenylacetic acid 26 revealed formation of proansamitocin (2). One possible explanation could be the oxidation of 26 to the corresponding 2‐oxo derivative which is followed by oxidative decarboxylation (like pyruvate→acetyl‐CoA) to AHBA 1 being activated as CoA thioester. The other carboxylic acids 27–31 did not act as processible mutasynthons in this study.With these results in hand we carried out large‐scale fermentations (up to 3.6 L) choosing the most promising benzoic acids 12, 14–16 and 25. The selection was based on semiquantitative analysis of signal intensities in the UPLC chromatograms (for details see the Supporting Information). Consequently, we did not include aminobenzoic acids 13 as well as aminobenzyl 23 and hydroxymethylbenzoic acids 24 for large‐scale fermentations despite the fact that test fermentations revealed PKS processing.As a result we isolated new proansamitocin derivatives 32–38 after purification by column chromatography (SiO2), size exclusion chromatography (LH‐20) and preparative HPLC (C18 and CN columns). Most remarkably, supplementation of 3‐hydroxy‐5‐(hydroxymethyl)benzoic acid (25) to a growing culture of strain AH1 yielded three products that resulted from complete PKS‐processing. Besides the seco‐acid 38, the two epimeric, ring‐expanded macrolactones 37 a and 37 b, were collected (Figure 5). Support for their formation was gained when comparing the chemical shift (δ) of the benzylic methylene protons of benzyl alcohols 25 and 38 with the epimeric macrolactones 37 a,b [ArCH
2O‐: 25 ([D6]DMSO) δ=4.46 (s, 2H) ppm; 37 a and 37 b ([D6]benzene) δ=5.05 (d, J=13.0 Hz, 1H) and 4.97/4.92 (d, J=13.0 Hz, 1H); 38
δ=4.50 (s, 2H) ppm]. These downfield shifts are diagnostic for O‐acylated secondary alcohols as found in macrolactones 37 a,b. It has to be noted that proansamitocins are prone to epimerization at C10, evident from the isolation of proansamitocins 2 and 4 that we have never noted for fully processed ansamitocin P3 (3).7
Figure 5
Fermentations using mutasynthons 12, 14–16 and 25 as well as isolation of proansamitocin derivatives 32–36, macrolactones 37 a and 37 b and seco‐acid 38.
Fermentations using mutasynthons 12, 14–16 and 25 as well as isolation of proansamitocin derivatives 32–36, macrolactones 37 a and 37 b and seco‐acid 38.The two 10‐epimers 37 a,b were identified by comparison of the coupling constants J
10,11 and J
10,12 with those found in epimeric proanamsitocins 2 and 4 (Figure 3 for major 10R isomer 37 a: J
10,11=7.0 Hz and J
10,12=0.9 Hz; for minor 10S isomer 37 b: J
10,11=5.3 Hz and J
10,12=1.7 Hz). It has to be stressed that this is the first example of an ansamitocin derivative ever reported that bears an expanded ring system and a lactone instead of a lactam substructure.The monitoring of the mutasynthetic experiment that leads to macrolactones 37 a,b revealed that seco‐acid 38 is not the precursor of the macrolactones or a side product of macrocyclization but rather its hydrolysis product (Figure 6). Over a period of 12 days the titers for macrolactones 37 first increased to about 5 mg/L before it a dramatically decreased until it could not be detected anymore. During that time, the formation of seco‐acid 38 continuously increased to a value of about 4 mg/L. It must be noted that formation of the corresponding macrolactones using A. pretiosum (ΔAHBA, strain HGF073) was not found in previous experiments.16
Figure 6
Monitoring the formation of macrolactone 37 (•) and seco‐acid 38 (○) by AH1 over a period of 12 days (feeding of 1.65 mmol/L on days 3, 4 and 5 p.i.); calibration was carried out with 2 as reference (see SI).
Monitoring the formation of macrolactone 37 (•) and seco‐acid 38 (○) by AH1 over a period of 12 days (feeding of 1.65 mmol/L on days 3, 4 and 5 p.i.); calibration was carried out with 2 as reference (see SI).In summary, expanding the concept of mutasynthesis by designing multiblocked mutant strains of Actinosynnema pretiosum opens the door to access new derivatives of biosynthetically advanced macrocyclic PKS intermediates. For the first time a ring expanded 20‐membered ansamitocinmacrolactone could be generated, revealing an unusual promiscuity and chemoselectivity of the amide synthase asm9 that commonly catalyzes macrolactam formation. These results suggest that this amide synthase may become a chemical tool of interest for performing macrocylizations.11
Conflict of interest
The authors declare no conflict of interest.As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.SupplementaryClick here for additional data file.
Authors: Michael J Smanski; Hui Zhou; Jan Claesen; Ben Shen; Michael A Fischbach; Christopher A Voigt Journal: Nat Rev Microbiol Date: 2016-03 Impact factor: 60.633